JPH0475182A - Data driving type information processor - Google Patents

Abstract

PURPOSE:To easily perform a high-functional arithmetic processing and to facilitate the description of a program, as well by providing a high-functional instruction processing means which stores processing information required for the processing of high-functional instructions previously and performs arithmetic processing for a data packet having the high-functional instructions. CONSTITUTION:When the data packet including high-functional instructions is inputted to an arithmetic processing means 106, the high-functional instruction processing means 109 performs the arithmetic processing for the data packet according to the previously stored processing information required for the execution of the high-functional instructions. Consequently, even when the high- functional instructions which require much processing information, e.g. instructions correlation arithmetic, classification arithmetic, and calculation arithmetic of structure data are executed, the structure of packets of data inputted to the arithmetic processing means 106 can be equalized to the structure of data packets including simple instructions. Further, the device can be utilized efficiently and the labor required for program preparation is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data-driven information processing device, and particularly to a data-driven information processing device in which a plurality of instructions are driven simultaneously.

[Prior Art] In a conventional Neumann type computer, various instructions are stored in a program memory in advance as a program, and the instructions are sequentially read out by sequentially specifying addresses in the program memory by a program counter, and the instructions are executed. be done.

On the other hand, a data-driven information processing device is a type of non-Neumann type computer that does not have the concept of sequential execution of instructions using a program counter. Such a data-driven information processing device employs an architecture based on parallel processing of instructions. In data-driven information processing devices, instructions can be executed as soon as the data to be operated on is available, and multiple instructions are simultaneously driven by the data, so programs are executed in parallel according to the natural flow of data. be done. As a result, the time required for calculations is considered to be significantly reduced.

FIG. 35 is a block diagram showing an example of the configuration of a conventional data-driven information processing device. FIG. 36 is a diagram showing an example of a data flow program stored in the program storage unit in the data-driven information processing device shown in FIG. 35. FIGS. 37AIm to 37D are diagrams showing an example of a field configuration of a data packet processed in the data-driven information processing device of FIG. 35. Hereinafter, a conventional data-driven information processing device will be described with reference to FIGS. 35 to 37D.

A data packet as shown in FIG. 37A is applied to the input section 101 from the outside. As shown in FIG. 37A, this data packet contains a small number of nodes (both node numbers and data). The data packet input from the input unit 101 is given to the program storage unit 102.Program storage unit 102
stores a data flow program as shown in FIG. Each line of this data flow program consists of a node number, an instruction code, and a flag. The program storage unit 102 reads out the node number, instruction code, and flag in a predetermined line of the data flow program by addressing based on the node number of the input data packet, as shown in FIG. 1
It stores it in a new data packet along with the data included in the data packet input from 01, and outputs the data packet. FIG. 37B shows a data packet output from the program storage unit 102.

The data pair detection unit 103 performs a waiting process, that is, a firing process, for data packets output from the program storage unit 102. That is, when the instruction code indicates a two-manual instruction, two different data packets having the same node number are detected, and data A and B contained in those data packets are processed as shown in FIG. 37C. It is stored in one data packet and output. Note that when the instruction code of the input data packet indicates a one-man command, the paired data detection unit 103 outputs the input data packet as is.

The arithmetic processing unit 104 performs arithmetic processing based on the instruction code on the data packet output from the paired data detection unit 103, and stores data C indicating the result of the calculation in the data packet as shown in FIG. 37D. and output it.

The output unit 105 outputs the data packet to the outside or returns it to the input unit 101 and inputs the data again based on whether a flag included in the data packet received from the arithmetic processing unit 104 is set. Controls whether the information is circulated within the drive type information processing device.

FIG. 38 is a diagram showing an example of a data flow graph. In FIG. 38, node N1 indicates an addition instruction, node N2 indicates a multiplication instruction, and node N3 indicates a subtraction instruction. Further, node N4 indicates a decrement instruction, and node N5 indicates an increment instruction. Node Nl, N
2. The command at node N3 is a two-man command, and the command at node N4. N5
This command is a one-person command. The calculation result of node N1 is referenced by nodes N2 and N3.

[Problems to be Solved by the Invention] In the conventional data-driven information processing device as described above, the arithmetic processing unit 104 performs arithmetic operations.

It was configured to perform only basic operations such as logical operations. This is because when the arithmetic processing unit 104 attempts to execute a complex operation that requires a large number of parameters, the parameters necessary to execute such operation must be carried in a data packet. This is because the data packet length would be extremely long (and the configuration of the device would become extremely large. There are operations (relational operations, classification operations, etc.) for manipulating data groups (such as groups of data that come together to form a single meaning).

As an example, let us consider how to accomplish the structure data combination operation shown in FIG. 39 using the conventional data-driven information processing device shown in FIG. 35. 39th
The operation shown in the figure is an operation in which when the second term of structure data A and the first term of structure data B match, they are combined to generate structure data C. Note that the first term of structure data C contains the value of the first term of structure data A, and the second term of structure data C contains the value of the second term of structure data A. The second term of structure data B is entered into the third term of data C.

FIG. 40 is a diagram showing the operational steps required when the conventional data-driven processing device shown in FIG. 35 attempts to perform the structure data manipulation shown in FIG. 39. That is, in order to accomplish the operation shown in FIG. 39 with a conventional data-driven information processing device,
By analyzing the contents of the operations shown in the figure in detail, breaking them down into a plurality of basic operations, and executing each basic operation one by one, the overall process shown in Figure 39 can be achieved. is required.

As mentioned above, in conventional data-driven information processing devices,
For example, in order to perform processing that requires a large number of parameters, such as manipulating structured data, an extremely large number of basic calculation execution steps are required, which poses the problem of not being able to utilize the equipment efficiently. . In addition, when creating a program, such complex processing must be broken down into basic operations and written, which results in the problem of requiring a high degree of skill and a great deal of effort to create the program. there were.

This invention was made to solve the above-mentioned problems, and even high-performance calculations that require a large number of parameters can be executed in the same way as ordinary basic calculations. It is an object of the present invention to provide a data-driven information processing device in which the program can be easily written.

[Means for Solving the Problems] A data-driven information processing device according to the present invention includes an input means, a program storage means, a paired data detection means, an arithmetic processing means, and an output control means.

The input means inputs a data packet including at least a node number and data. The program storage means stores the data flow program, reads out at least the next node number and instruction information from the data flow program based on the node number included in the data packet given from the input means, and reads out at least the next node number and instruction information from the data flow program, and A new data packet containing the number, command information and data in the data packet is generated and output. The pair data detection means receives the data packet output from the program storage means, detects two data packets having the same node number to form a pair, and detects a new data packet having the data in both of these data packets. Generate and output. The arithmetic processing means receives the data packet output from the paired data detection means, performs arithmetic processing on the data included in the data packet based on command information included in the data packet, and includes data indicating the result of the arithmetic operation. Generate and output data packets. The output control means receives a data packet output from the arithmetic processing means and controls whether the data packet is output to the outside or to the input means. The arithmetic processing means includes branching means, simple instruction processing means, and high-performance instruction processing means.

The branching means branches the data packet received from the program storage means into a data packet having a simple instruction and a data packet having a sophisticated instruction. The simple instruction processing means performs arithmetic processing on a data packet having a simple instruction among the data packets branched by the branching means. The high-performance instruction processing means stores processing information necessary for executing the high-performance instruction in advance, and performs arithmetic processing on a data packet having a high-performance instruction among the data packets branched by the branching means. Be informed.

[Operation] In this invention, the high-performance instruction processing means in the arithmetic processing means stores in advance the processing information necessary for executing the high-performance instruction, and based on this processing information, the high-performance instruction processing means included in the data packet is processed. In order to execute a functional instruction, even when executing a highly functional instruction (for example, a relational operation instruction) that requires a large amount of processing information, the structure of the data packet input to the arithmetic processing means is the same as that of the simple instruction. This is similar to a packet, and there is no need to carry a large amount of processing information from the program storage means to the arithmetic processing means in a data packet.

Therefore, the arithmetic processing means can perform a wide variety of processing with greater flexibility than in the past.

[Embodiment] FIG. 1 is a schematic block diagram showing the configuration of an embodiment of the present invention. In the figure, in the embodiment of FIG. 1, the arithmetic processing section 1 is replaced with the arithmetic processing section 104 in FIG.
06 is provided. The rest of the configuration is similar to that of the conventional data-driven information processing device shown in FIG.

The arithmetic processing section 106 includes a branching section 107 , a simple instruction processing section 108 , a high-performance instruction processing section 109 , and a merging section 110 . The branching unit 107 is connected to the paired data detection unit 103.
The input data packet is branched to a simple instruction processing unit 108 and a high-performance instruction processing unit 109 based on the instruction code included in the data packet given from the input data packet. That is, if the instruction code included in the input data packet is a simple instruction (an instruction that instructs a basic operation such as an arithmetic operation or a logical operation), the branching unit 107 sends the data packet to the simple instruction processing unit. 108. On the other hand, if the instruction code included in the input data packet is a high-performance instruction (for example, an instruction that instructs a high-performance operation that requires a large number of processing parameters, such as a relational operation), the data packet is It is output to the instruction processing unit 109. The simple instruction processing unit 108 executes basic operations such as arithmetic operations and logical operations, and its configuration may be similar to the configuration of the arithmetic processing unit 104 in FIG. 36. Therefore, the simple instruction processing unit 108 outputs data indicating the calculation result of the simple instruction to the merging unit 110 in a data packet. The high-performance instruction processing unit 109 stores processing information necessary for executing high-performance instructions, and executes the high-performance instructions based on this processing information.

A controlled device 200 is connected to the high-performance command processing section 109 . The high-performance command processing unit 109 processes various processes that occur in the process of executing high-performance commands on this controlled device 2.
Do this for 00. Examples of the controlled device 200 include a data memory, a communication device, and a display. Further, the high-performance instruction processing unit 109 outputs data indicating the calculation result of the high-performance instruction to the merging unit 110 in a data packet. The merging unit 110 merges data packets output from the simple instruction processing unit 108 and the high-performance instruction processing unit 109 and outputs the data packets to the output unit 105.

FIG. 2 is a block diagram showing a more detailed configuration of the high-performance instruction processing section 109 shown in FIG. 1.

In the figure, the high-performance instruction processing unit 109 includes a node number extraction circuit 109a, an instruction code extraction circuit 109b, an input data extraction circuit 109C, and a specification data memory 1.
09d, processing circuit 109e, and output packet combining unit 1
09f. The data packet input from the branching unit 107 is given to a node number extraction circuit 109a, an instruction code extraction circuit 109b, and an input data extraction circuit 109c, and the node number, instruction code, and data are extracted in each of them. The node number extracted by the node number extraction circuit 109a is stored in the spec data memory 1.
Given on 09d. The spec data memory 109d stores in advance processing information (parameters, spec data, or program information) necessary for high-performance commands as shown in FIG. 3, for example. The spec data memory 109d uses the node number given from the node number extraction circuit 109a as an address, and reads and outputs processing information for the high-performance instruction corresponding to the node number. The processing information read from the spec data memory 109d is given to the processing circuit 109e. In addition, processing circuit 1
09e contains the instruction code extracted by the instruction code extraction circuit 109b and the input card data extraction circuit 109C.
The data extracted by is given. Processing circuit 10
9e executes a high-performance instruction corresponding to the given instruction code based on the processing information given from the spec data memory 109d. During this execution process, the processing circuit 109e controls the controlled device 200. For example, if the controlled device 200 is a data memory,
Controls rewriting of internal storage information. Furthermore, the processing circuit 109e outputs data indicating the result of the arithmetic processing of the high-performance instruction to the output packet synthesis unit 109f. The output packet synthesis unit 109f puts the data given from the processing circuit 109e into a data packet and outputs it.

The output of the output packet combining section 109f is given to the merging section 110 in FIG.

Next, in the embodiment shown in FIGS. 1 and 2, a description will be given of how a combination operation of structure data as shown in FIG. 39, for example, is processed. A join operation as shown in FIG. 39 is treated as one of the relational operations without being decomposed into basic operations. That is, in the data flow program in the program storage unit 102, a join instruction is simply written in a corresponding node to instruct a join operation. Therefore, the data packet output from the program storage unit 102 includes a joi instruction code as shown in FIG.
This will include n instructions.

Further, the data packet does not carry the actual data of each structure data, but carries pointer information, that is, address information of the structure data.

The paired data detection unit 103 detects the same node number (for example, N
), the two sets of data packets are queued. As a result of this waiting, if two sets of data packets having the same node number N are detected, the paired data detection unit 10
3 performs firing processing as shown in FIG. 4, and outputs pointer information of structure data A and pointer information of structure data B in one data packet. Branch part 10
7, the instruction code included in the data packet given from the paired data detection unit 103 is, for example, j.

If it is a high-performance instruction such as an in instruction, the data packet is given to the high-performance instruction processing unit 109 . The high-performance instruction processing unit 109 extracts the node number, instruction code, and data from the given data packet. The extracted node number N is stored in the spec data memory 109d.
The processing information corresponding to this node number N is read from the spec data memory 109d. As shown in FIG. 3, the spec data memory 109d stores various information necessary for processing high-performance instructions for each node number. This processing information includes the formats of input data and output data, processing parameters, and the like.

The data read from the spec data memory 109d is given to the processing circuit 109e. The processing circuit 109e is constituted by a digital signal processor, a microcomputer, etc., and executes high-performance instructions based on the processing information provided. That is, the processing circuit 109e extracts the structure data A and B in the controlled device 200 (here, data memory) based on the pointer information of each structure data A and B extracted by the input data extraction circuit 109c. , and the second term of structure data A and structure data B
It is determined whether or not the second term is equal to the second term. If it is determined that both data are equal, structure data C is written to a certain area within the data memory 200. At this time, the value of the first term of structure data A is written to the first term of structure data C, the value of the second term of structure data A is written to the second term of structure data C, and the value of the second term of structure data A is written to the second term of structure data C. The second term of structure data B is written into the third term of structure data C. In this way, the controlled device 2
In 00, structure data C, which is the result of combining structure data A and B, is generated. on the other hand,
The processing circuit 109e provides pointer information of the structure data C to the output packet synthesis unit 109f. The output packet synthesis unit 1゜9f generates a data packet containing pointer information of the structure data C as shown in FIG.
Output to 10.

As described above, in the embodiments shown in FIGS. 1 and 2, data packets input to the arithmetic processing unit 106 are also used when executing high-performance instructions that require a large amount of processing information, such as relational operation instructions. The structure of is similar to the data packet of a simple instruction as shown in FIG. 4, and there is no need to carry a large amount of processing information from the program storage unit 102 to the arithmetic processing unit 106 in a data packet. Therefore, the arithmetic processing unit 106 can perform a variety of processes with greater flexibility than conventional data-driven information processing devices. Also, when writing data flow programs,
Since highly functional calculations can be handled in the same way as basic calculations, program creation becomes extremely simple.

Note that there are various types of relational operations executed in the embodiments shown in FIGS. 1 and 2 in addition to the join operation shown in FIG. 39. Below are some other examples of relational operations.

(1) Cartesian Product Operation (Product) The Cartesian product operation is an operation that combines and connects all elements of structure data A and structure B, as shown in FIG. 5A.

(2) Restriction operation (restriction) As shown in FIG. 5B, a restriction operation is an operation for extracting an element of a specific item from a certain structure data as a key.

(3) Union operation (union) As shown in FIG. This is an operation that concatenates .

(4) Projection operation (projection) As shown in FIG. 5D, a projection operation is an operation that extracts a specific item from a certain structure data to generate new structure data.

All of the above relational operations are written in the data flow program in the program storage unit 102 as high-performance instructions.
It is executed in the high-performance instruction processing unit 109.

Next, a program development apparatus constructed using the data-driven information processing apparatus according to the present invention will be described. The program development device described below enables the user to define a single specification from multiple perspectives using multiple languages, and provides a software development environment with excellent understandability, descriptive ability, and operability. It is a completely new program development device developed for the purpose of converting a software system into an executable program and executing it, thereby making it possible to simulate and execute the generated program. Please point it out in advance (.

FIG. 6 is a diagram schematically showing the overall configuration of a program development device configured using the data-driven information processing device of the present invention. The program development device shown in FIG. 6 includes editors E1 to En, each of which has drawing and editing functions according to different expression formats, and a main unit 1 that creates a program according to the specification description given by the user. include.

Each of the graphic editors E1 to En is preferably realized by an independent process, and each of the graphic editors E1 to En can communicate with the main device 1. Here, figure editor E1~
The communication between each En and the main device 1 is as follows.
This is because, in consideration of ease of program development and future functional expansion, the processes of the graphic editors E1 to En are connected to the main unit 1 through interprocess communication using, for example, sockets.

The representation formats provided by the graphic editors E1 to En include a graphic representation format and a non-procedural description format.

The graphic representation format includes representation formats such as a functional block diagram and a sequence chart.

As shown in FIG. 7, a functional block diagram is an expression format that expresses connection relationships between modules (units that have a sense of unity in software).

FIG. 7 is a diagram showing an example of a specification description format using a functional block diagram. In FIG. 7, the external module 10
A data connection relationship between the internal module 11 and the internal module 11 is shown. The external module 10 is a module that expresses an external function that is not subject to specification description. The internal module 11 is a module that expresses a function to be specified. A data arc 12 defines the flow of data between the external module 10 and the internal module 11 together with the input/output ports P1 to P4.

This internal module includes primitives and parts (partially completed specification descriptions, registered in files).

The sequence chart is a diagram showing the causal relationship of input/output data, the relationship of data transmission/reception between modules, etc., as shown in FIG. In FIG. 8, the external module 10
3 shows the flow of data between the internal module 11 and the internal module 11. In this sequence chart, a module and its port are indicated by a single order line (vertical line). The sequence line indicates the time axis and also represents the temporal relationship in which signals flow. The data flow between this module 10.11 is represented by signal line 13. This signal line 13 indicates data connections and causal relationships between modules.

Non-procedural description formats include expression formats shown in tabular formats such as relational tables and decision tables. The relational table shows the relational data structure as shown in FIG. In Figure 9, the relationship table is
An area 14 that displays the name of the table, an item name area 15 that displays the names of the items included in this table, and the data type of the item (int (integer), float (floating point)
and String), and an area 17 in which the actual data value of the item is input/output. The decision table, as shown in FIG. 10, is an expression format representing the selection structure of processing. In Figure 10, the decision table is
An area 18 for displaying the decision table module name, an area 19 for displaying judgment conditions, an area 20 for displaying the name of the process to be executed as a result of the judgment, an area 21 for displaying the judgment for the conditions, and a condition judgment area. It includes an area 22 that displays a judgment indicating whether or not to execute processing based on the above.

In addition to the above-mentioned representation formats, available representation formats include data block diagrams showing data inclusion relationships, and table operation diagrams showing structure data processing centered on relational operations.

FIG. 11 shows an example of the representation format using a data block diagram, and FIG. 12 shows an example of the representation format of the table manipulation diagram.

In FIG. 11, the data block diagram includes a data set name area 23, an area 24 that displays the number of elements included in the data in this area 23, and member data names or subordinate data that constitute the members of the data set in the area 23. An area 26 that displays the data set name of the hierarchy and the data type of member data (int, floatSSirtngs 5t
ruct). This area 24
is written only when the data has an array structure. When the data type is 5 truct, this data indicates that the data is hierarchical with respect to the lower level. The connection relationship of each data is indicated by an AND connection 27 or an OR connection 28.

The table operation diagram shown in FIG. 12 shows, as an example, a state in which relational tables B1 and B2 are processed according to a predetermined operation to form a new relational table B3. This operation is called a merger operation. The operations in the table operation diagram are not limited to this merging operation, but include various relational operations, classification operations, calculation operations, and the like. Relational operations are operations that manipulate tables to satisfy certain relationships. The classification operation is an operation of rearranging the rows of a table according to a predetermined order, such as alphabetical order or numerical order.

A calculation operation is an operation to obtain a calculation result such as a total value or an average value for a specific column of a table.

Returning to FIG. 6, the main device 1 fuses information obtained from the shortcut specification descriptions generated by the user using the graphic editors E1 to En to create construct information that includes control information necessary for program execution. and a mutual conversion device Cv, which mutually converts information between graphic specification descriptions having different expression formats and generates basic information between module hierarchies and presents them to the user.
An execution device EX that interprets and executes a program according to the construct information generated by the program and presents the execution results to the user, graphic editors E1 to En for describing graphic specifications, mutual conversion devices CV1, an execution device EX, and a parts management device CP. Input/output device 1 that manages communication for exchanging information with
0. and an integrated file management device UF that integrally manages data structures handled by each processing device EX, CV, and CP.

The component management device CP registers a partially completed specification description in software as a "component" and manages operations for reusing the registered "component".

The construct information is information obtained by converting the generated specification description information so as to correspond to the operation method of the processing model including the control structure. A data-driven model is used as an example of this processing model. In order to use this data-driven model as a processing model, it is necessary to generate control nodes and control arcs. In the construct information, nodes are associated with modules of specification description information, and arcs are associated with data connections between modules.

The input/output device (2) 0 has the functions of an interface with the user, execution control of each processing device based on instructions from the user, and management of data exchanged between the processing devices.

The interface with the user includes the following:

■Description of specifications in an expression format using a graphic editor,■
Display of mutual conversion results, ■ Correspondence is a function, ■ Citation function, ■
Prototyping function, ■Parts registration/reuse, and ■
Contains documentation output.

The following methods can be used to describe specifications in each expression format using a graphic editor.

(a) The user specifies the type of expression format and the name of the specification description. This specification description name is specified when reusing an already registered specification description, and when creating a new specification description, a new specification description name is input.

(b) Specify graphical elements and expression formats in the existing specification description in order to make the already written specification more detailed or to proceed with the description in a hierarchical manner.

In this case, specifications are written using an expression format corresponding to the designated graphical element.

Specifications can be written for the same module in different expression formats. In this case, the mutual conversion device C2 informs the editor of the corresponding expression format according to the internal mutual conversion rule table search, and this corresponding converted expression format is displayed to the user. Since one specification description can be executed using different expression formats, it becomes easier to understand the description contents from multiple aspects, and it becomes possible to create a more complete specification.

The association function is a function that associates semantically equivalent information (for example, data or functional modules) between different expression formats. This function is required to interrelate various types of independently written specification information in order to sequentially integrate specifications written according to various expression formats.

That is, as shown in FIG. 13, the module M7 expressed in a functional block diagram can also be expressed using an operation table diagram. Module M represented by this table manipulation diagram
7 and M7 shown in the functional block diagram are semantically equivalent. In this case, the modules M7 expressed in different expression formats are related to each other as being semantically the same, as indicated by the arrows in the figure. This operation is called a "matching" operation.

Furthermore, each expression format may share information with each other. In this case, the amount of work during writing can be reduced by quoting the already written specification information when writing in another expression format.

This use of information written in one expression format in another expression format is called "quotation." For example, as shown in FIG. 13, the module M10 represented by a relational table is the same as the table MIO represented by a table operation diagram. In this case, as shown by the arrow in the figure, the table MIO of the relational table is cited when describing in accordance with the expression format in the table operation diagram. Similarly, table M8 in the table manipulation diagram is the same as table M8 expressed as a relational table.

In this case as well, M8 expressed in the relational table is cited when writing in accordance with the expression format of the table manipulation diagram.

The prototyping function is a function for checking the execution results of partially completed specification descriptions. In this prototyping, the execution results of the specification description can be confirmed independently for each layer. The user specifies input/output data and creates input data during prototyping according to the specified data structure.

The interface for parts registration/reuse instructs the parts management device CP to register parts and use the registered parts.

The document output function outputs a compact specification description written in each expression format to a printer as a hard copy for specification descriptions written using various expression formats, and can be used as a software design document as is. possible.

The interface between the input/output device IO and each processing device will be explained. First, the interface with the graphic editor will be explained.

The input/output device IO controls the startup and termination of various graphic editors designated by the user. It also manages information required when starting and terminating the editor, such as the name of the input specification description and the type of graphic editor. This operation includes:

(a) When the user specifies the type of expression format and the name of the specification description, the input/output device ■0 starts the graphic editor corresponding to the specified expression format, and at the same time inputs the name of the specification description. Pass to this activated shape editor.

(b) Each graphic editor is provided with a function for starting other graphic editors. The input/output device is notified of the graphical element (graphical element included in the operation tool displayed by the editor) specified in an editor and the type of expression format. The type and shape of the symbol of this graphic element differs depending on the type of expression format. The input/output device (2) 0 activates a graphic editor corresponding to the specified expression format and sends the specified graphic element to the activated graphic editor.

(c) Termination control of specification description. Each graphic editor is equipped with a termination operation tool. This termination operation tool notifies the input/output device IO of termination of the graphic editor. In response to this termination notification, the input/output device IO saves information such as the name of the created specification description and the type of editor, and then terminates the graphic editor that notified the termination.

Input/output device 0 also provides an interface with a graphics editor and interconverter. When describing specifications using various expression formats, or when using multiple specification descriptions mutually, the input/output device IO performs the following processing with the graphic editor and the mutual conversion device CV.

(a) The drawing information described by the user is sequentially sent to the mutual conversion device Cv via the input/output device IO at the stage when each graphic element is described. The mutual conversion device is this input/output device■0
After performing a predetermined process (this will be described later) according to the information given from the input/output device 0, the data is returned to the input/output device 0.

In this case, the returned data is specification information that can be expressed in a different expression format corresponding to the input specification description information. The data returned from the mutual conversion device C2 is attached with an identifier ID of the specification description of the return destination together with the processing result. The input/output device 10 uses this identifier ID.
The specification description corresponding to is specified, and the processing result formed by the mutual conversion device Cv is returned to the destination graphic editor.

Further, when the mutual conversion device C2 detects a contradiction in the input specification description, the input/output device 0 presents a message indicating the detected contradiction to the user.

(b) In the specification writing process, the input/output device IO determines whether or not the two graphic elements specified by the user on the two graphic editors can be logically correlated. .

If the correspondence is possible, the input/output device IO returns the same logical specification information to both graphic editors. If the correspondence is not possible, a message indicating this inability is presented to the user.

(c) Quotation When quoting specification information between multiple graphic editors, the input/output device ■0 transfers the drawing information and specification information of the graphic element specified by the user on the graphic editor to the quotation light graphic editor. .

Next, the interface between the input/output device IO and the execution device EX will be explained. Input data information and position information of input/output data required during execution (for example, prototyping) of the execution device EX are transmitted from the graphic editor to the input/output device IO in the same manner as when writing specifications. The input/output device IO uses an execution handler (
(not shown). When this preprocessing is completed, the input/output device IO starts the execution device EX in response to an execution start instruction from the user. The execution result data is transferred from the above-mentioned execution handler to the input/output device ■0
is displayed on the display device DP via. This display device D
The display on P is executed, for example, by activating an output result display routine.

The interface between the input/output device IO and the component management device CP includes the following.

(a) Interface when registering parts When the specification description created by the user is partially completed, the specification description information and part name of the registration target specified by the user are sent to the parts management device cp via the input/output device IO. transmitted to.

(b) Interface when reusing parts The part name specified by the user on the graphic editor is transmitted to the parts management device CP via the input/output device 10. The component management device CP searches for component information corresponding to the given component name via the file management device UF, and returns the searched component information to the activated graphic editor via the input/output device IO.

If the user wants to refer to parts information, input/output device I
When O starts the parts specification display routine based on instructions from the user, the parts information expressed approximately during this time is displayed on the display device DP.

The functions of mutual conversion device C2 will be explained. This mutual conversion device CV generates an effective executable program by integrating the contents of specification descriptions written by users using various compaction expression formats, and also converts specific aspects of a given specification description ( It has the function of converting information acquired from one representation format or hierarchy into another representation format. Here, the executable program is a program obtained by converting construct information having a structure dependent on a processing model (data-driven model) so as to match the execution method of the virtual machine.

FIG. 14 shows an example of the structure of the specification description information integrated by the mutual conversion device C■. This specification description information includes module information MI, sequence information Sl, and data structure information DI.
including.

The module information Mf is information regarding the operations of individual modules constituting this software system and data connections between modules. This module information is
The hierarchical relationship between each module is generated based on the specification description using a functional block diagram. This module information also integrates data structure information through data connection relationships between modules. Module information is functional module,
Contains information regarding the decision table module and structure data manipulation module, respectively. The information of each module corresponds to processing contents that can be defined in each expression format of a functional block diagram, a decision table, and a table operation diagram expressed by the graphic editor. Each module will be explained below.

(A) Functional module: A functional module is information about internal submodules that constitute a module and information about data connection relationships between each submodule. Attributes included in the submodule information correspond to descriptive elements provided by the functional block diagram, and include primitives, parts, internal modules, files, branches, connections, and the like. The hierarchical structure of modules is saved by attaching module IDs (identifiers) attached to modules whose submodules define detailed information.

Data connection relationships between modules are associated with arcs in the functional block diagram. This arc represents the relationship between the data producing submodule and the data consuming submodule. The data structure of data is managed by referring to data structure information using a data set ID.

(B) The decision table module is information that indicates information regarding selection processing based on information obtained from the decision table, by means of condition determination and processing executed based on the determination result.

The relationship between the content of each condition determination and the corresponding process to be executed is managed by the event ID.

(C) Structure manipulation module information is information that expresses information regarding structure data manipulation based on information obtained from the table manipulation diagram using data to be manipulated and operations that act on each data. The correspondence between the data to be operated and each operation is managed by the data ID and operation ID. The data structure of each data is Dataset I
It is managed by D by referring to data structure information.

Data structure information is information regarding the definition of the type and structure of data consumed/produced by the software system. The hierarchical relationship of the data structure is generated based on the description using the data block diagram. This data structure information is
It includes inclusive data structure information, relational data structure information, etc., and includes atom information as data without structure.

(A) Inclusive data information is information indicating the structure of data in an inclusive/exclusive relationship. This included data information is
For each component data, the hierarchical relationship of the data structures is preserved by having a data set ID given to the data structure that defines the detailed structure.

(B) The relational data structure is information indicating the basic data type of each item data that constitutes the relational data.

(C) Atom information is information indicating the basic data type (int, float, string) regarding atom data.

The sequence information includes attributes of the sequence line and information on signal lines input to and output from the sequence line. In this sequence information, since it is possible to define a plurality of sequences for one module, these relationships are managed using sequence IDs.

(A) Sequence line information is information indicating attributes related to vertical lines representing submodules in the sequence chart. The attributes of the sequence line include internal modules, files, etc., and each corresponds to a descriptive element of the sequence chart. Correspondence with submodules in module information is managed by module sub-ID.

(B) Signal line information is information indicating data connections between sequential lines (hereinafter referred to as signal lines) in the sequence chart. This signal line information also indicates the causal relationship between input and output data by simultaneously managing information on a group of output sequences that are connected to the input of the signal line and output. Correspondence with data dependence in module information is managed by arc ID.

As described above, the relationships between each piece of information constituting the specification description information are all managed by the identifier ID. As mentioned above, this identifier ID includes a dataset ID that identifies data structure information, a module ID that identifies module information, a module sub-ID that identifies internal sub-modules that make up the module, and a data connection between internal sub-modules. Contains the arc ID etc. Each time new information is added, the mutual conversion device Cv adds a new identifier ID.
Generate and manage.

As shown in FIG. 14, in the structure of the main description information, the specification content described using multiple graphic editors is hierarchical, with data reference relationships 32, module hierarchical relationships 33, and common information for the same module. Definition information relationship 3
Contains 4. Therefore, by using these relationships as a basis, specification description information for the entire target software system can be generated from partial specification descriptions defined in a plurality of different expression formats.

Specification description information that integrates each piece of information obtained using these individual expression formats is managed by the identifier ID as described above.

Examples of the data structure of specification description information obtained by integrating information obtained from such individual expression formats are shown in FIGS. 15 to 20.

Figure 15 shows the data structure in the specification description, Figure 16 shows the structure of sequential line information, and Figure 17 shows the structure of signal line information.
FIG. 18 shows the structure of relational data, FIG. 19 shows the structure of included data, and FIG. 20 shows the structure of atoms. In the structure of the specification description information shown in FIGS. 15 to 20, the presence or absence of correspondence, the presence or absence of change, etc. are determined by referring to the identifier ID attached to each piece of information.

FIG. 21 is a flow diagram showing the operation of the mutual conversion device.

Hereinafter, the operation of this mutual conversion device CV will be briefly explained with reference to the operation flow diagram of FIG. 21. Generation of specification description information is executed based on communication with the editor via the input/output device 10.

In step S1, the editor or input/output device IO
Data is given from. This data is specification description information, end command information, specification description creation start information, etc. If it is determined that the given information is an end command (step S2), the mutual conversion device C2 performs desired processing such as closing the file, and then ends its operation.

In cases other than the end command, the mutual conversion device CV generates necessary specification description information according to the given data (step S4). The generated specification description information is used to access the file management device UF, and the contents of the specification description information file are updated with the generated specification description information (step S5).

On the other hand, the mutual conversion device C2 generates construct information from the given specification description information (step S6). Regarding this generated construct information, by accessing the file management device UF again, the original construct information file is updated depending on whether the newly generated construct information is added or changed.

Next, this mutual conversion device CV searches for conversion rules stored in advance in the form of a table. By searching this conversion rule table, it is determined whether there is data that has the same meaning as the input data. That is, it is analyzed whether information given from a certain editor can be converted into an object in another compact representation format (for example, the correspondence between modules in a functional block diagram and vertical lines in a sequence chart) (step S9). If an object that can be converted to another representation format is retrieved through this analysis, convert it into a corresponding convertible intermediate object (step 510), and indicate the converted intermediate object obtained as a result of this conversion. Send the data to the corresponding editor via input/output device ■0 (step 511)
. The editor receiving the transmission at this time is an editor that can express this convertible object.

When all of this conversion processing is completed (step 512),
The mutual conversion device CV again waits for a semantically coherent object to be sent from the editor or the input/output device.

In step S9, if there is no convertible object, the mutual conversion device rILCv notifies the input/output device 0 to that effect and enters a state of waiting for data from the editor or the input/output device IO.

Here, when the result data written by the user is input via the input/output device IO, if one semantically grouped object is updated, the one grouped object is mutually It is sent to the conversion device Cv.

Furthermore, when a transformable space object is searched for in step S9, and if there are a plurality of transformable space objects, a graphic editor is activated that provides a representation format corresponding to each of the plurality of space objects. Ru. The graphical elements in the expression format provided by the simultaneously activated graphical editors are simultaneously displayed on the display device DP, for example, by a multi-window.

Next, an example of information exchange between the graphic editor and the mutual conversion device Cv will be described with reference to FIGS. 22A and 22B. In the example shown in FIGS. 22A and 22B, a graphic editor whose representation format is a functional block diagram is activated, and the mutual conversion device C generates a specification description based on the representation format of the functional block diagram. is shown as an example.

In response to activation of a module definition or data structure definition editor by a user via the input/output device IO, the mutual conversion device Cv generates new module information or data structure information. The type of module or data to be generated is determined from the type of graphic editor activated. At the same time, the mutual conversion device C2 generates a module ID or data set ID for the generated module information or data structure information and transmits it to the activated graphic editor. In FIG. 22A, the module ID is transmitted to the graphic editor E. In FIG.

In figure editor E, when the user adds a new internal module on that editor, this figure editor E
transmits the undefined module sub-ID together with the transmitted module ID and information indicating the attributes of the module to the mutual conversion device CV.

The mutual conversion device CV generates an internal module in response to the given undefined module sub-ID, and also generates a module sub-ID for identifying this new sub-module information and transmits it to the graphic editor E. do.

In the graphic editor E, when information regarding an existing internal module is changed or deleted in the functional block diagram, this change or deletion information is transmitted to the mutual conversion device Cv together with the corresponding module sub-ID. In response to this information, the mutual converter CV changes or deletes the corresponding submodule information.

When a plurality of internal modules are generated, arcs indicating data connections between the internal modules are written on the graphic editor E. In response, the graphic editor E inputs the module sub-ID that identifies the module that sends the data, the port ID that indicates the data output port of this internal module, the destination module sub-ID that inputs the data, and the module sub-ID that this destination module uses to input the data. The destination boat ID identifying the receiving boat and the undefined arc ID are transmitted to the mutual conversion device Cv. The mutual conversion device CV generates an arc in the internal module in response to this given information, and adds predetermined information to the given undefined arc D in order to identify this arc. I
It is transmitted as D to the graphic editor E. Next, when the data structure for the arc to be identified is specified by this arc ID, the graphic editor transmits the data set ID indicating this specified data structure together with the corresponding arc ID to the mutual conversion device CV. The mutual conversion device Cv registers the data set ID according to this given information. Next, the operation when generating a hierarchical module will be described with reference to FIG. 22B.

For existing internal modules in the functional block diagram,
When the user starts another graphic editor to define the details of this existing internal module, the graphic editor that started it responds by adding a new module ID and module sub-ID. A lower module ID (undefined) for identifying the module is transmitted to the conversion device C■. The conversion device CV generates new module information in response to this given information, and adds the module ID of the upper module at that time to the generated module (lower module). Thereby, the lower module ID is determined. The module ID of this generated module (lower module) is transmitted to the editor activated to define its details, and is also transmitted to the editor that initiated this activation. After this, when the description for the lower module is completed, the activated graphic editor will display the module ID along with information to indicate it.
, the module sub ID and the lower module ID are transmitted to the mutual conversion device CV.

In response to this information, the conversion device CV generates coupling information between the upper module and the lower module.

From the graphics editor on the activation side, information on the module ID corresponding to the lower module is sent to the mutual conversion device CV in accordance with the internal module information change sequence. Thereby, the hierarchical structure of module information is realized as a reference relationship from both upper and lower sides using this identifier ID.

The reference relationship between data and data structure information in data dependence of operations is realized as follows. When a correspondence relationship between an arc and a data structure description is established through a corresponding operation in a plurality of graphic editors, the dataset ID acquired at that time is transmitted to the mutual conversion device Cv according to the change sequence of data dependency information. Ru. This saves the structure of input/output data in the module, making it possible to extract the access sequence and lock range to the structure data.

As described above, all the specification description information constituting each module is managed by managing the identifier ID, thereby managing the relationship between each part information.

An example of generating a specification description of the entire software system using this hierarchical representation will be explained.

FIG. 23 is a diagram showing an example of creating more accurate module information from this hierarchical structure information. As shown in FIG. 23, in the specification description 41 of the upper layer, detailed description 42 of the lower layer of each module corresponds to the content expressed as the data transmission/reception relationship between module FOO and module FOI. and 43 and associating them in the interconversion device Cv,
The structure in which module FOO merges data a and data b to derive data C, while module FOI receives data C, outputs data d, and sends this output d back to module FOO is analyzed. Ru. This makes it possible to derive a repetition structure 44 in which input/output data b and data d of the functional module are the same data, and data d is repeatedly used again.

The functional module P controls the operation of the functional module representing "merge" and the rTF gate that transmits data as is; the "authenticity determination gate". The details of this control become clear by using the lower modules for the functional module P.

FIG. 24 is a diagram showing an example of a configuration for generating construct information from a combination of reduction expression formats (specification description) performed in this mutual conversion device. In the specification diagram 45 expressed as a functional block diagram, module P
It is not possible to determine whether the operation content of S and S is a selection structure that simply performs data copying.

However, by associating the descriptions 50 and 51 of the decision table with the respective modules P and S, the module P selects one of the modules Fl to F3 according to the magnitude relationship of the input data x and y. The branching module S is controlled to control the branching module S.
In accordance with the data from module P, a structure for transmitting data b to any of modules Fl to F3 is determined. This determines the selection structure of the module P, and further enables generation of control information for branch control in the branch module S. This makes it possible to generate construct information 46 that includes control information for the processing model.

The above-described embodiment shows a case where information for generating a construct is obtained by integrating a plurality of expression formats. However, as shown in FIGS. 25 and 26, it is also possible to reversely generate (mutually convert) information in individual expression formats from the integrated information.

FIG. 25 is a diagram illustrating mutual conversion between a functional block diagram and a sequence chart. In Figure 25,
Internal module information 6 obtained from the functional block diagram 60
1 is converted into ordinal line information 62, and an ordinal line on the sequence chart is generated according to this ordinal line information. Here, the conversion from internal module information to sequence line information is performed in the mutual conversion device CV as described above, and this information is transmitted to a graphic editor whose representation format is a sequence chart, and this graphic editor is activated. , a sequence line on the sequence chart is generated on the display device.

In the following explanation as well, information is exchanged between this mutual conversion device and a graphic editor that expresses functional block diagrams and sequence charts.

Further, in the sequence chart 63, an internal module M2 is further added to the order line information 62 obtained from the sequence chart 60. This sequence line information indicating the newly generated internal module is converted to internal module information on the functional block diagram, and the internal module is generated on the functional block diagram. At this time, signal line information (signal lines A-D) obtained by forming signal lines in the sequence chart is converted into data dependency information.

As a result, an arc corresponding to the generated signal line is generated in the functional block diagram 64. This provides data flow between each module. However, since the signal line information only indicates the flow of data and does not indicate which ports of the module are connected, the boat connections are undefined in this newly generated functional block diagram 65. A display 68 indicating this is given to the user. The user looks at this and defines the boat.

In FIG. 25, an initial state 62 of the sequence chart is generated from information 60 obtained from the functional block diagram, and descriptions are added to the sequence chart 62 in the initial state, that is, generation of internal modules and data flow. A new functional block diagram 65 is obtained by generating and adding the contents of the functional block diagram according to the added sequence chart description 63. In this case, in the newly formed functional block diagram, by presenting the incomplete part 68 in the functional block diagram where the boat is undefined to the user, it is possible to prompt the user to write new information, making it more accurate. This makes it possible to construct a software system that is easy to use.

Furthermore, by using mutual conversion processing, lower layer information in the hierarchical description can be generated.

In the specification description shown in FIG. 26, a functional block diagram 72 is formed from the information expressed in the sequence chart 71, and a lower module 73 for module M1 and a lower module 74 for module M2 in this functional block diagram 72 are formed. are doing. When a new sequence chart 75 is given to this mutual conversion configuration, that is, when a structure in which the module M1 receives data A from an external module and sends out data D is described, the mutual conversion device CV performs the following: The information obtained in this sequence chart 75 is added to the already given module 72. In this case, by detecting a match between the data ID and the data A, the module M
1 is recognized to receive data A and derive data B and D. From this structure, by considering the lower module 73 of the hierarchical representation, Ml has a branch structure in its lower structure, and data B and data D
The sub-modules Mll and Ml2 respectively derive
is detected to contain. In this case, since the relationship between data B and data C in module M2 is preserved as in the sequence chart 71, its lower structure 78 has the same representation as the lower module 74.

Furthermore, in contrast to the sequence chart 71 shown in FIG. 26, a new sequence chart 79 is added as shown in FIG.
is formed, a functional block diagram is obtained according to the information obtained by the formed sequence chart. In this case, by detecting the match of the data ID with respect to data B, if data A and data D have an exclusive data structure, it is determined that the lower module Mll that receives data A and the lower module Mll that receives data M2 have an exclusive data structure for module M1. A merge structure 80 is generated, which includes a module M12 and a lower structure that derives data B by merging the outputs of the modules Mll and M12.

Next, the functions of the execution device EX will be explained. Execution device E
X includes a conversion unit that converts the construct information generated by the mutual conversion device Cv into an executable program according to the operation method of an execution model (virtual machine), and an execution unit that executes plot typing. Note that the execution section is constituted by a data-driven information processing device as shown in FIG. However, the high-performance command processing unit 109 in FIG.
It is set in.

The executable format information obtained by converting the construct information includes connection information that expresses the flow of processing represented by the data flow and the structure of each process, and number links (various IDs) that appear in this connection information. It is broadly divided into constant information, data structure information, and file information. In this execution format information as well, the connection information is grouped for each layer, and 20 node numbers are uniquely generated while maintaining the relationship in which corresponding boats are linked between layers. When the execution unit in the execution unit EX executes a predetermined specification, it tracks the connections of nodes and executes the program in sequence. Therefore, conversion into an information format suitable for this execution means conversion into an information format in which connection relationships between nodes are indicated by arc information attached to this node information. Next, the operation of this execution device EX will be explained with reference to FIG. 28, which is an operation flow diagram. In FIG. 28, the operation flow of the execution device EX is displayed using a sequence chart.

The user specifies a data arc into which data on the specification description is to be input. A graphic editor corresponding to this specified data arc is activated, an input window is opened, and displayed on the display device DP. A template corresponding to the data structure of the specified data arc is displayed in the input window. Input data is set by filling this template. Setting the output data is the same as setting the input data, and when you specify the data arc whose data you want to monitor in the specification description, an output window is opened and a template corresponding to the data structure of the specified data arc is displayed.

The input data specified by the user is input to the input/output device 10.
The input data is written to the integrated file management device UF via the mutual conversion device Cv, and the input data is updated. When the data update management is completed in the file management device UF, information indicating the completion is transmitted to the input/output device 10 via the mutual conversion device C2. When the user instructs the start of prototyping execution, this execution start instruction is sent to the input/output device ■0
and is given to the execution unit EX via. The execution device EX is
The integrated file management device UF refers to the area in which the input data set in the input window is written, reads this input data, and executes the program corresponding to the specified specification description.

The execution device EX reads the input data from the integrated file management device UF, sequentially executes the functions set on the nodes while tracking the connections of the nodes included in the execution format information, and checks whether the output data is specified in the arc. Get the output data when When the execution device EX obtains the output data, it writes the output data to the integrated file management device UF. This updates the output data. When the integrated file management device notifies the execution device EX of the completion of this data update, the execution device EX then notifies the input/output device IO of the completion of prototyping.

When the input/output device IO receives this execution completion, the conversion device Cv
Notify the end of execution. In response to the completion of this execution, the mutual conversion device CV refers to the output data written in the integrated file management device UF, reads out the output data, and transmits it to the input/output device IO.

Input/output device ■0 displays the given output data on the display device DP
Display in the output window that was opened earlier.

As mentioned above, simulations targeting any hierarchy in the specification description can be executed by simply setting the input data and output data format, so the created specifications can be checked and reviewed. . Next, the functions of the parts management device CP will be explained.

A complete specification in a specification description is called a "part". The component management device CP that manages this component has the following functions.

■Parts information presentation function, ■Parts name list presentation function, ■
They are: presentation of table item data type, ■ presentation of construct information, ■ component specification display function, ■ component registration function, and 0-component extension function. This component management device CP can be accessed by the input/output device IO, the mutual conversion device CV, and the execution device EX.

Next, referring to FIG. 29, the component reference operation will be explained.

FIG. 29 is a diagram showing a sequence chart of a component reference operation flow. If the user wishes to refer to a component, the user selects "reference component" from the menu provided by the input/output device IO. In response to this "component reference" instruction, the input/output device IO requests a component name list from the component management device CP in order to display a list of registered components. The parts management device IfCP searches the parts information file via the file management device UF, and transmits the list of registered component names to the input/output device ■0 via the parts management device CP. The input/output device 10 activates a graphic editor, opens a component reference window (reference WD), and displays a list of registered components on the display device.

The user uses a data input device such as a "mouse" or a character input device to specify a component whose input/output information is desired to be referenced from among the components shown in the component reference window (reference WD).

In response to the reference part designation from the user, the input/output device IO activates a parts specification editor (specification ED) and hands over the designated part name to this parts specification editor.

The component specification editor (specification ED) refers to the component specification file stored in the specification file and displays this component specification. As a result, the user can view the part specification at the specification description level expressed in a shorthand manner on the display device DP.

Next, the component quoting operation will be explained.

When quoting a component while writing a specification, select "Component" in the drawing mode of the functional block diagram editor or sequence chart editor.

An editor quoting a component issues a request for component request information to the input/output device IO. The input/output device IO opens a window for quoting parts and displays a list of registered parts in response to an instruction from the user in the same manner as in the part reference operation described above.

The user specifies which part he/she wishes to quote from among the parts shown in this part reference window. The input/output device IO sends the component information for this specified component to the component management device CP.
request. The parts management device CP is a file management device U.
The component information file is searched through F, and component information for the designated component is provided to the input/output device IO.

The input/output device 10 further provides the component information provided by the component management device CP to an editor that wishes to quote this component. The editor draws based on the supplied part information and notifies the input/output device IO of changes in the specification description.

Next, the functions of the integrated file management device UF will be explained. Information regarding the contents of the specification description handled by this development device is expressed as one integrated file. The function of manipulating this file is realized by the integrated file management device UF as a data structure manipulation command (high-performance command). The integrated file management device UF is connected to other processing devices (input/output device IO1 mutual conversion device C, execution device It has an interface that is declaratively coupled with EX and parts management device CP, and integrally manages internal data structures expressing specification description information.

Other processing devices that use the integrated file management device UF are:
Pass the operation code for the required data structure manipulation to the integrated file management device UF. The integrated file management device UF stores information (
For example, information as shown in FIG. 3) is stored in advance, and based on the information, the designated data structure operation is executed, and the execution result is returned to the processing device that called the integrated file management device UF. That is, the integrated file management device UF has the function of the high-performance command processing unit 109 as shown in FIG.

As data structures used in the program development apparatus of the present invention, a scalar (atom) and a list structure are employed. If you use this list structure,
This is because structures such as "array", "record", and "vector" can be expressed. When using this list structure, ■ accessing elements by subscript (identifier),
and ■ Access to the element by key matching is performed.

In this list structure, the logical identifier rfid for the identifier (name) of the entire data structure to be processed is
J, and a logical identifier [didJ for identifying an array in which one element is a data area of the size of the array element indicated by rsizej in the data structure identified by [fidJ]. By using these identifiers, the access area in the file is determined.

When the integrated file management device UF is referenced using the scalar data structure, reference and update are performed using the data name.

Each processing device operates in parallel while exchanging information with each other. Therefore, multiple access requests are generated to the integrated file management device UF. Since these accesses occur dynamically, the processing device that uses the integrated file management device UF cannot determine whether multiple accesses are occurring. Furthermore, even if it is possible to determine that multiple accesses have occurred, it is difficult to determine whether the processing device that requested the access later can access the file separately. The processing device cannot know. For this reason, the integrated file management device UF includes a mechanism for guaranteeing consistency for a series of accesses regarding files in a multi-process environment. As such a mechanism, the side using the integrated file management bag ff1fUF makes an access request, and then determines whether to subsequently request access to the same record of the same file and what kind of access request to make. By notifying the management device UF, the integrated file management device UF
A lock control mechanism is provided for locking a necessary range of files.

Lock control is implemented as one primitive. There are two types of files managed by the integrated file management device UF: temporary files and permanent files. Temporary files are consumed once they are used. Permanent files are saved before and after the target program is executed. When executing primitives that use files, a temporary file is always created to write to. Next, a specific example of creating a software system using a program development apparatus according to an embodiment of the present invention will be described.

Now, let's consider the case of writing a module to search by "key" in an inventory management problem. Now, as a specific problem, I will describe the following problem.

The search data is a list of items. This product list has the structure of "Product list: 2 product code, product name, quantity, and delivered light." The search item (key) is the product name code or quantity. For searches using quantity as the "key",
List the items corresponding to the given quantity or more of the item code.

First, as shown in FIG. 30, the interface with the outside is described. The system to be described is a product name search module 91 that inputs a product name code and outputs a corresponding list.
It includes an a1 item list file 91b1, an external module (user) 91c, and a quantity search module 91d that inputs quantities and outputs a corresponding list. This description from a functional aspect is expressed by a functional block diagram 91.

At this time, the module information expressed in this functional block diagram 91 is converted into a sequence chart and displayed (see the lower right part of FIG. 30).

Next, as shown in FIG. 31, the order relationship of data flow between each module will be described. The first sequence chart 92 describes a product name search sequence, and the second sequence chart describes a quantity search sequence. This sequence chart 92 and 93
The signal line described in is immediately converted to a data arc on the functional block diagram 94 and displayed. At this time, since the boat information is not reflected in the sequence chart, the boat information will be described in the functional block diagram 94.

Here, although the relationship between data is described using a sequence chart, the relationship between data may also be described using a functional block diagram. By mutually converting and reflecting common information between different expression formats in this way, inconsistencies between descriptions can be avoided.

Next, the data structure will be described as shown in FIG. A relevant list (not shown) and a product list are described in a relational table 95, and product name codes and quantities are described in a data block diagram 96. In the example shown in FIG. 32, only the minimum required product name code and quantity are described in the applicable list. The described data structure, the data arc on the functional block diagram, and the signal line on the sequence chart are correlated by a matching operation. By using this association operation, complicated naming work is relieved.

Next, as shown in FIG. 33, the details of the internal module will be defined. In FIG. 33, the product name search module 91a included in the modules to be described is a file operation, so it is described using a table operation diagram 96. In this table operation diagram 96, the input/output information has already been converted and reflected according to the functional block diagram described earlier. The structure of the table data in this table operation diagram 96 is quoted from the relational tables 95a and 95b. The cited data structure is associated with the original relational table. Therefore, if the original relational table data structure is changed, the data structure in this table operation diagram 96 is also changed. This can prevent forgetting to make changes due to changes in specifications.

By describing the product name search module using the table operation diagram shown in FIG. 33, its specification description is completed and becomes executable. For this executable product name search module, a third
As shown in Figure 4, we perform prototyping to verify the specifications. That is, the product code of the product list 96a is used as input data, and the corresponding list 96b is used as output data.

The product name code included in the product list is entered in the product code of the product list 96a. Thereafter, by the execution of the apparatus, the item corresponding to this item name code 3 is displayed on the corresponding list along with its corresponding quantity. According to the execution results obtained through this prototyping, the specification description contents are changed in accordance with the target specification, and then executed again to verify.

After this, when the description of the specifications of the search system is completed, this is one partially completed module and is registered as a "component".

According to the program development device described above, one specification definition information (one function) can be described using multiple expression formats, so a software system can be defined from multiple aspects, and high-quality software can be created. A system can be constructed.

[Effects of the Invention] As described above, according to the present invention, when a data packet containing a high-performance instruction is input to the arithmetic processing means, the high-performance instruction processing means executes the high-performance instruction stored in advance. Since the calculation processing of the data packet is performed based on the processing information necessary for When executing the data, the structure of the data packet input to the arithmetic processing means is the same as that of a data packet containing a simple instruction, and a large amount of processing information is transferred from the program storage means to the arithmetic processing means in a data packet. This will avoid situations where you would have to carry it. As a result, not only can equipment be used efficiently, but when writing a program, complex arithmetic processing can be handled at the same level as normal simple instructions without having to be broken down into simple instructions. Labor can be significantly reduced compared to conventional data-driven information processing devices.

[Brief explanation of the drawing]

FIG. 1 is a schematic block diagram showing the configuration of an embodiment of the present invention. FIG. 2 is a block diagram showing a more detailed configuration of the high-performance instruction processing section in FIG. 1. FIG. 3 is a diagram showing an example of processing information for high-performance instructions stored in the spec data memory in FIG. 2. FIG. 4 shows that in the embodiment shown in FIGS. 1 and 2,
FIG. 3 is a diagram showing a process in which a high-performance command is processed along the flow of data packets. FIGS. 5A to 5D are diagrams illustrating typical relational operations that can be executed by the high-performance instruction processing unit. FIG. 6 is a diagram schematically showing the overall configuration of a program development device configured using the data-driven information processing device according to the present invention. FIG. 7 is a diagram illustrating a functional block diagram that is one of the specification description expression formats used in the program development apparatus shown in FIG. 6. FIG. 8 is a diagram illustrating a sequence chart that is one of the specification description expression formats that can be used in the program development apparatus shown in FIG. FIG. 9 is a diagram illustrating a relationship table that is one of the specification description expression formats used in the program development apparatus shown in FIG. 6. FIG. 10 is a diagram illustrating a decision table that is one of the specification description expression formats that can be used in the program development apparatus shown in FIG. FIG. 11 is a diagram illustrating a data block diagram that is one of the specification description expression formats that can be used in the program development apparatus shown in FIG. FIG. 12 is a diagram illustrating a behind-the-scenes diagram that is one of the specification description expression formats that can be used in the program development apparatus shown in FIG. FIG. 13 is a diagram illustrating "quote" and "corresponding material" operations when a specification is described using a plurality of expression formats. FIG. 14 is a diagram showing an example of the configuration of specification description information generated by the program development apparatus shown in FIG. 6. FIG. 15 is a diagram showing a list of the configurations of specification description information used in the program development apparatus shown in FIG. 6. FIG. 1.6 is a diagram showing the structure of sequence line information included in specification description information. FIG. 17 is a diagram showing the structure of the signal line information shown in FIG. 15. FIG. 18 is a diagram showing the structure of the relational data shown in FIG. 15. FIG. 19 is a diagram showing the structure of the included data shown in FIG. 15. FIG. 20 is a diagram showing the structure of the atom shown in FIG. 15. FIG. 21 is a flow diagram showing the operation of the mutual conversion device in the program development device shown in FIG. FIGS. 22A and 22B are diagrams illustrating a data transmission/reception sequence between the graphic editor, which is a specification description expression means, and the mutual conversion device. FIG. 23 is a diagram illustrating a manner in which construct information is generated from a combination of a plurality of expression formats. FIG. 24 is a diagram illustrating another example of generating construct information from a combination of a plurality of expression formats. FIG. 25 is a diagram illustrating a mutual conversion mode of a functional block diagram and a sequence chart executed in the mutual conversion device. FIG. 26 is a diagram illustrating an example of a mode in which lower layer information in the hierarchical description of sequence specification information is generated using a mutual conversion function between a sequence chart and a functional block diagram. FIG. 27 is a diagram illustrating another example of generating lower layer information in the hierarchical description of the specification description using the mutual conversion function. FIG. 28 is a sequence chart showing the prototyping execution operation in the program development apparatus shown in FIG. 6. FIG. 29 is a sequence chart diagram showing a component reference operation in the program development apparatus shown in FIG. 6. FIGS. 30 to 34 are diagrams illustrating a specific example of specification description using the program development apparatus shown in FIG. 6. FIG. 35 is a schematic block diagram showing the configuration of a conventional data-driven information processing device. FIG. 36 is a diagram showing an example of a data flow program stored in the program storage section shown in FIG. 35. FIGS. 37A to 37D are diagrams showing field configurations of data packets processed in the conventional data-driven information processing apparatus shown in FIG. 35. FIG. 38 is a diagram showing an example of a data flow graph. FIG. 39 is an illustrative diagram showing an example of a structure data combination operation. FIG. 40 is a flow diagram showing operational steps required when the conventional data-driven information processing apparatus shown in FIG. 35 performs the process shown in FIG. 39. In the figure, 1 is the main device, IO is the input/output device, CV is the mutual conversion device, EX is the execution device, CP is the parts management device,
UF is an integrated file management device, E. E1 to En are graphic editors, DP is a display device, 101 is an input section, 102 is a program storage section, 103 is a paired data detection section, 105 is an output section, 106 is an arithmetic processing section, 107
108 is a simple instruction processing section, 109 is a high-performance instruction processing section, 110 is a confluence section, 109a is a node number extraction circuit, 109b is an instruction code extraction circuit, 109C is an input data extraction circuit, 109d is a spec data memory ,1
09e indicates a processing circuit, and 109f indicates an output packet combining section.

Claims (1)

[Claims] (1) An input means for inputting a data packet including at least a node number and data, storing a data flow program, and inputting the data flow program based on the node number included in the data packet given from the input means. a program storage means for reading out at least a next node number and instruction information from the program storage means, and generating and outputting a new data packet including those node numbers, instruction information and data in the data packet; Receive the output data packet,
Paired data detection means for detecting two data packets having the same node number to form a pair, and generating and outputting a new data packet having data in both of these data packets; arithmetic processing means that receives a data packet, performs arithmetic processing on the data contained in the data packet based on command information contained in the data packet, and generates and outputs a data packet having data indicating the result of the calculation; and output control means for receiving a data packet output from the arithmetic processing means and controlling whether to output the data packet to the outside or to the input means, the arithmetic processing means comprising: the program storage means branching means for branching a data packet received from a data packet into a data packet having a simple instruction and a data packet having a high-function instruction; and a data packet having the simple instruction among the data packets branched by the branching means. simple instruction processing means that performs arithmetic processing; and processing information necessary for processing high-function instructions is stored in advance, and among the data packets branched by the branching means, a data packet having the high-function instruction a data-driven information processing device, the data-driven information processing device comprising: a high-performance instruction processing unit that performs arithmetic processing based on the processing information;